Protocols & Rails — AI Ad Integrity
Monetize AI without losing trust.
No Stamp → No Sponsored Serve
When AI generates or serves sponsored content, "we reviewed it" is not a gate. Good Proof verifies sponsored AI influence at serve time with a live status check and portable evidence.
- Gate sponsored serve by Status Link: not VALID → block or escalate
- Proof that travels: counterparties verify scope • expiry • signer • status by link
- Evidence Pack (IDA format): reusable across audits, disputes, and partner reviews
Fail-closed•Append-only•Scope-bounded
Scope-limited verification. Not certification. Acceptance depends on counterparty or programme requirements.
What this rail is
Runtime verification for sponsored AI influence at serve time.
What this rail is not
- • Ad quality scoring
- • Certification
- • Claim-truth adjudication
- • Replacement for legal review or brand-safety measurement
Good Proof does not score ad quality. It verifies whether sponsored AI influence met defined controls at decision time.
Why this rail, now
AI-generated sponsored content is scaling across search, social, and conversational surfaces.
The failure mode isn't "the ad was wrong." It's: a sponsored output served and nobody can prove it was authorized, disclosed, and within policy at serve time.
Screenshots and logs don't satisfy regulators, brand safety partners, or litigation — they don't travel, aren't revocable, and don't capture policy state.
Pressure points converging:
- •Trust: Users can't distinguish AI-generated sponsored content from organic
- •Disclosure: Obligations vary by jurisdiction. This rail provides portable evidence that configured controls were applied at serve time.
- •Procurement: Brand safety partners and advertisers demand verifiable controls
Good Proof provides a portable verification object: a Status Link + IDA Evidence Pack that proves what was authorized, disclosed, and within policy at serve time.
Why ad-tech buyers are moving now
Sponsored influence scrutiny, disclosure requirements, and partner assurance demands are converging.
Sponsored influence scrutiny increasing
AI-generated sponsored content is scaling across search, social, and conversational surfaces — regulators and partners are watching.
Disclosure expectations tightening
Jurisdictional disclosure requirements vary and are expanding. Portable proof of disclosure state at serve time is becoming a procurement requirement.
Partner and advertiser assurance gaps
Brand safety partners and advertisers demand verifiable controls — screenshots and internal dashboards don't travel.
Disputes over takedown/change history
Sponsors challenge takedowns. Teams can't prove what was authorized vs. what changed. Status history solves this.
Cross-system evidence fragmentation
Evidence lives in ad servers, review queues, policy dashboards, and email threads. None of it is portable or revocable.
Demand for data-minimized verification
Partners need proof without user profiles, targeting data, or raw creative payloads. References and hashes satisfy this.
Good Proof provides scope-limited verification evidence and stop-rely semantics. It is not a certification.
What a Stamp proves (and what it doesn't)
Proves (within lane scope)
- Ad action class + outcome
- Instruction/serve timestamp
- Sponsor identity reference
- Disclosure state at serve time
- Policy/model version binding
- Validity state (VALID / NEEDS_REFRESH / WITHDRAWN / NOT_VERIFIED)
Does NOT prove
- Ad effectiveness or CTR
- Claim truthfulness
- Raw targeting/user profile data
- Creative quality
- Certification or regulatory compliance
In disputes: Status Link = reliance state now. IDA Evidence Pack = fileable serve-time snapshot.
Scope-limited verification. Not certification. Acceptance depends on counterparty or programme requirements.
What gets stamped in AI Ad Integrity lanes
High-impact ad action classes (define per programme):
Sponsor identity reference
Reference to verified sponsor identity at ad instruction time (not full KYC payload).
Creative provenance hash
Hash/reference for creative provenance chain (not creative asset storage by default).
Claim-risk class
Risk classification of claims in the ad content (health, financial, political, etc.).
Disclosure payload reference
Reference to machine-readable disclosure: sponsor, material connection, and regulatory flags.
Audience/jurisdiction constraint refs
Segment/jurisdiction IDs for targeting boundaries (not raw user profile data).
Policy/model version IDs
Version references for ad policy, content moderation model, and enforcement rules.
If sponsored output cannot be safely recalled, it must be verifiable and revocable before serve.
What you get (two artefacts, one standard)
Status Link (authoritative now)
A counterparty-verifiable link that returns current validity within scope.
- Returns: status, scope, expiry, verified_at, signer, verify_url
- Fail-closed: unreachable = NOT_VERIFIED
- Built for ad pipelines, exchange flows, and automated gates
IDA Evidence Pack (snapshot then)
View full details →A time-stamped snapshot you can forward, file, and cite.
- Built for audits, disputes, partner reviews, and regulatory inquiries
- Minimal disclosure by default (proof ≠ payloads)
- Programme-configured redaction matrix
PDFs are great for filing. Status Links keep them current.
Runtime status enforcement
Every sponsored output must pass a status gate before serve. Non-VALID states trigger fail-closed behavior.
VALID
Ad verified. Serve permitted.
NEEDS REFRESH
Policy/creative changed. Re-verify.
WITHDRAWN
Stop serving immediately.
NOT VERIFIED
No proof exists. Block.
Fail-closed: if status ≠ VALID at serve time, sponsored output does not serve.
VALID = valid within scope (not a guarantee of claim truthfulness or ad effectiveness).
Serve-time behavior (revenue-safe pattern)
VALID
Serve sponsored output.
NOT VERIFIED / NEEDS REFRESH / WITHDRAWN
Do not serve sponsored output.
Optional fallback
Serve eligible non-sponsored output and log state + timestamp for audit.
When status changes — and what it means
Status triggers define when a Status Link moves to NEEDS_REFRESH or WITHDRAWN. Understanding these ensures fail-closed enforcement at serve time.
NEEDS_REFRESH
When any of these occur, re-verify before serve.
Sponsor identity/account status change
Creative updated or modified post-approval
Policy version change (ad standards, content rules)
Claim-risk classification updated
Jurisdiction/audience targeting rules changed
Model version update affecting content scoring
Disclosure requirement change (regulatory update)
Third-party verification vendor change
NEEDS_REFRESH means "re-verify before you serve," not "schedule a meeting."
WITHDRAWN
Stop-serve signal. Ad must not serve.
Sponsor account suspended or terminated
Creative flagged for policy violation (confirmed)
Regulatory takedown order received
Materially misleading output confirmed under policy/regulatory determination
Sponsor identity compromise suspected
Material misrepresentation discovered
Court or regulator mandates stop-serve
Fail-closed: Wherever the Status Link is checked, if WITHDRAWN → block or escalate.
Integration touchpoints
1
Pre-serve gate
Insert status check at the final sponsored output decision point. If status ≠ VALID, block sponsored serve.
2
Programmatic / exchange compatibility
Status check before serve in exchange flows. Structured fail-closed response handling for bid/serve pipelines.
3
Telemetry hooks
Policy, sponsor, and context change events sync status updates. Material changes trigger NEEDS_REFRESH automatically.
4
Verify API integration
Millisecond status checks for serve-time gates. Structured fail-closed response with scope + expiry + signer.
High-impact gating only. Everything else runs normally.
AI-Agent Era
AI-agent era controls
Prompts can drift. Ad policy enforcement must not.
Material change in sponsor/policy/model/configNEEDS_REFRESH
Integrity or policy boundary breachWITHDRAWN
Timeout/unreachable verification routeNOT_VERIFIED (fail-closed)
Exception lane requiring human finalityGuardian path (optional)
Good Proof does not decide ad outcomes; it controls whether sponsored actions are safe to serve.
What's inside the IDA Evidence Pack
Serve-time snapshot for audit, disputes, and regulatory inquiries.
Ad summary + lane scope boundary
Instruction/serve timestamp + evidence window
Sponsor identity reference (abstracted)
Creative provenance chain (hash, not content)
Claim-risk classification + policy version
Disclosure payload reference
Jurisdiction/targeting constraint refs
Verification transcript + timestamps
Redaction matrix (what's excluded by design)
Data-minimized by default. Programme-gated access when required.
What counterparties can verify without internal access
Current validity state: VALID / NEEDS REFRESH / WITHDRAWN / NOT VERIFIED
Scope boundaries and expiry window
Signer authority reference (system or Guardian when required)
Change triggers fired (refresh reason code) — without exposing payloads
Verification route and optional signed verify response (programme scoped)
Optional tamper-evident anchoring for high-assurance programmes
No login portal required. Evidence portability for disputes, procurement, and regulator review.
Who buys this in AI Ad Integrity
Platform, advertiser, and ecosystem buyers with high-impact ad accountability.
Platform + advertiser buyers
Trust & Safety
Pain: Ads get challenged and we can't prove what was approved at serve time.
Outcome: Status Link shows ad validity state; withdrawal propagates instantly.
Book an Ad Integrity SprintLegal / Policy / Regulatory Affairs
Pain: Regulatory scrutiny on AI-generated ads; no portable proof of disclosure.
Outcome: Evidence Pack captures disclosure + policy version at instruction time.
Book an Ad Integrity SprintAdvertiser Operations / Monetization
Pain: Sponsors dispute takedowns; no way to show what changed.
Outcome: Status transitions are traceable; refresh triggers document policy changes.
Book an Ad Integrity SprintPrivacy / Data Governance
Pain: Verification evidence exposes too much — targeting data, user profiles, creative content.
Outcome: Data-minimized by default. References and hashes, not raw payloads.
Book an Ad Integrity SprintProduct / Ads Engineering
Pain: Adding controls means rearchitecting the ad pipeline.
Outcome: Status gate at final serve decision point. Verify API returns in milliseconds.
Book an Ad Integrity SprintProgrammatic / Exchange Operations
Pain: Exchange partners need proof without system integration or portal access.
Outcome: Verify by link. Status + scope + expiry returned without internal account access.
Book an Ad Integrity SprintExternal verifiers
Regulators
Pain: Enforcement investigations require disclosure state evidence that doesn't exist portably.
Outcome: Fileable Evidence Pack with serve-time snapshot and redaction matrix.
Auditors / Brand safety partners
Pain: Audit evidence is system-bound and stale by the time it's exported.
Outcome: Live Status Link for current state; Evidence Pack for time-of-serve record.
Legal counsel (advertiser/platform)
Pain: Dispute resolution lacks portable, time-stamped evidence of serve-time controls.
Outcome: Evidence Pack is structured for counsel review and regulatory filing.
Where budget comes from
Usually funded from existing trust, safety, and compliance lines — not new category spend.
Trust & safety budget
Trigger: Regulatory inquiry, partner audit, or disputed takedown
Why it fits: Portable evidence + fail-closed reliance control reduce dispute resolution cost.
Regulatory disclosure budget
Trigger: Jurisdictional disclosure requirement change or enforcement action
Why it fits: Serve-time disclosure proof satisfies evidence requirements without system exports.
Legal / dispute cost containment
Trigger: Sponsor disputes, litigation, or regulatory challenge
Why it fits: Decision-time snapshot + live status make serve-time controls defensible.
Advertiser assurance / partner compliance
Trigger: Enterprise partner requiring verification as procurement condition
Why it fits: Counterparty-verifiable status without portal access or NDA for basic verification.
Platform integrity / risk operations
Trigger: AI ad scaling outpacing review capacity or evidence portability
Why it fits: Machine-checkable serve-time gates reduce manual review dependency for status checks.
Monetization governance
Trigger: AI-generated ad flows entering production without portable verification
Why it fits: Pre-serve gate ensures every sponsored output has verifiable, revocable proof.
Start with one high-impact ad lane and prove dispute/audit friction reduction before expansion.
A Global Human Layer
Our Mind Chill Guardian Story
A global human layer that software can't fake.
When liability lands on a person, the sign-off should too.
Conflict-checked · Rotation-based · Audit-traceable · Programme-scoped
When Guardians are used (only when required)
Most ad decisions remain automated. Humans step in only where human finality is required: high-risk claim appeals, disputed takedowns, or edge cases that cannot be safely automated.
Guardians are exception-only, not required for normal real-time ad serving. Controls: conflict checks, rotation, multi-review for high-risk claims, sampling audits.
From calming minds to defending outcomes
Mind Chill began in 2017 as immersive art built to reduce anxiety and create calm at scale. Then the same feeds that buried calm and rewarded outrage started training the systems that now make real decisions. We didn't want more rhetoric. We wanted receipts.
The moment it clicked
A message arrived: someone's child felt safer because of what they experienced. Around the same time, lived experience inside our own community made one thing obvious: the nuance that matters in high-impact decisions can't be reliably reduced to a prompt. So we designed a human layer for the edge cases—structured, scope-bound, and auditable.
Guardians are not a "panel." They're a network.
Mind Chill Guardians come from different countries, backgrounds, and lived realities. That diversity is not branding—it's risk reduction. It makes decisions harder to game, easier to challenge, and more credible under scrutiny. Guardians do not "run the system." They review only what the lane requires humans to own.
Receipts over rhetoric
Operational Guardians plug into Good Proof lanes as a controlled finality mechanism: conflict checks, rotation, multi-review where required, and an audit trace tied to a Status Link. Minimal disclosure by default. If a decision is appealed months later, you can show what happened, within scope, without dumping sensitive payloads.
Why buyers choose Guardians
Lived experience at the edge cases (not a generic helpdesk)
Conflict-checked + rotation-based (anti-rubber-stamp by design)
Multi-review on high-risk lanes (when the programme requires it)
Audit-traceable outcomes (defensible in disputes, audits, procurement)
Minimal disclosure by default (proof, not payloads)
Not a certification. Scope-limited verification. Acceptance depends on counterparty/programme requirements.
How it works (simple)
1
Define the ad action class + scope
Sponsor types, claim classes, jurisdictions, evidence window, and what triggers refresh or withdrawal.
2
Require a pre-serve Status gate
No Stamp, NOT VERIFIED, or NEEDS REFRESH → block sponsored serve or escalate (per programme runbook).
3
Ship portable proof
Issue the Status Link and generate the Evidence Pack automatically at serve time.
4
Humans step in only when required
Guardians handle exceptions and disputes inside defined scope, with anti-rubber-stamp controls.
Procurement-ready clause
Template language for your legal team.
"For [High-Impact Ad Action Classes], Supplier shall issue a Good Proof Stamp prior to serve. Buyer may verify status via the Status Link. Stamps returning NOT VERIFIED, NEEDS REFRESH, or WITHDRAWN shall block or escalate per programme runbooks."
Anti-spoof controls (contract-grade)
- verify_url host MUST exactly match official verifier
- HTTPS/TLS required — no insecure overrides
- Redirects forbidden — any redirect = NOT_VERIFIED
Schedule A (template — programme terms)
Definitions + operating rules procurement teams can copy/paste.
1. Definitions
- High-Impact Ad Action Class: a sponsored or high-risk ad output designated for Stamp gating (e.g., sponsored serve, high-risk claim serve).
- Status Link: the verification endpoint returning validity state + scope boundaries.
- Evidence Window: the time period during which supporting materials are retained for audit/dispute.
- Evidence Pack: time-stamped snapshot for filing/disputes (IDA format).
- Scope Boundary: the defined limits of what a Stamp covers (action class, expiry, programme).
2. Required states + enforcement behavior
- VALID→ may serve within scope.
- NEEDS_REFRESH/ NOT_VERIFIED / WITHDRAWN → must block or escalate per runbooks.
- Fail-closed:timeout/unreachable ⇒ NOT_VERIFIED.
3. Withdrawal / stop-serve semantics
- WITHDRAWN is returned wherever the Status Link is checked.
- No sponsored serve may proceed on WITHDRAWN.
- Optional: programme hooks/notifications for stop-serve distribution.
4. SLA placeholders (complete per programme)
Verifier availability target: [___]%. P99 verify latency: [___] ms. Propagation time: [___] seconds. Support turnaround: [___] hours.
5. Retention defaults
Default evidence window: [90] days (configurable 30–365). Verification transcripts retained per programme schedule.
Not legal advice. Bracketed variables to be completed by the parties.
Procurement pack available: architecture summary, data handling overview, subprocessors, retention options.
30-day Sprint outcome
What you get in 30 days
One ad action class defined (e.g., sponsored output serve, high-risk claim)
Scope boundaries documented (sponsor types + claim classes + jurisdictions)
Pre-serve gate integrated
Refresh/withdraw triggers configured
Counterparty verification route tested
One redacted IDA Evidence Pack specimen generated
Go/no-go rollout recommendation
Due Diligence FAQs
No Stamp → No Serve.
Make sponsored AI outputs verifiable, scope-bound, and revocable by link.
Scope-limited verification. Not certification. Acceptance depends on counterparty or programme requirements.